Intrinsically motivated reinforcement learning for human-robot interaction in the real-world

TL;DR

This paper introduces an intrinsically motivated reinforcement learning framework enabling robots to learn human-like social skills from real-world interactions without explicit instructions, resulting in more human-like decision-making.

Contribution

It presents a novel intrinsic motivation-based reinforcement learning method that allows robots to acquire social skills from natural human-robot interactions in uncontrolled environments.

Findings

Robots learned social skills from real-world interactions.

Robots made more human-like decisions than those with explicit rewards.

The method improved social skill acquisition in uncontrolled settings.

Abstract

For a natural social human-robot interaction, it is essential for a robot to learn the human-like social skills. However, learning such skills is notoriously hard due to the limited availability of direct instructions from people to teach a robot. In this paper, we propose an intrinsically motivated reinforcement learning framework in which an agent gets the intrinsic motivation-based rewards through the action-conditional predictive model. By using the proposed method, the robot learned the social skills from the human-robot interaction experiences gathered in the real uncontrolled environments. The results indicate that the robot not only acquired human-like social skills but also took more human-like decisions, on a test dataset, than a robot which received direct rewards for the task achievement.